Interactions of two abscisic-acid induced WRKY genes in repressing gibberellin signaling in aleurone cells

Summary

Gibberellins (GA) promote while abscisic acid (ABA) inhibits seed germination and post-germination growth. To address the cross-talk of GA and ABA signaling, we studied two rice WRKY genes (OsWRKY51 and OsWRKY71) that are ABA-inducible and GA-repressible in embryos and aleurone cells. Over-expression of these two genes in aleurone cells specifically and synergistically represses induction of the ABA-repressible and GA-inducible Amy32b alpha-amylase promoter reporter construct (Amy32b-GUS) by GA or the GA-inducible transcriptional activator, GAMYB. The physical interactions of OsWRKY71 proteins themselves and that of OsWRKY71 and OsWRKY51 are revealed in the nuclei of aleurone cells using bimolecular fluorescence complementation (BiFC) assays. Although OsWRKY51 itself does not bind to the Amy32b promoter in vitro, it interacts with OsWRKY71 and enhances the binding affinity of OsWRKY71 to W boxes in the Amy32b promoter. The binding activity of OsWRKY71 is abolished by deleting the C-terminus containing the WRKY domain or substituting the key amino acids in the WRKY motif and the zinc finger region. However, two of these non-DNA-binding mutants are still able to repress GA induction by enhancing the binding affinity of the wild-type DNA-binding OsWRKY71 repressors. In contrast, the third non-DNA-binding mutant enhances GA induction of Amy32b-GUS, by interfering with the binding of the wild-type OsWRKY71 or the OsWRKY71/OsWRKY51 repressing complex. These data demonstrate the synergistic interaction of ABA-inducible WRKY genes in regulating GAMYB-mediated GA signaling in aleurone cells, thereby establishing a novel mechanism for ABA and GA signaling cross-talk.

Introduction

The aleurone layer of cereal grains serves as a convenient system for studying gibberellic acid (GA) and abscisic acid (ABA) signal transduction pathways. During germination of cereal grains, GA is secreted from embryos into aleurone cells to promote the expression of hydrolytic enzymes, such as α-amylases, which are necessary to utilize stored reserves within the endosperm for seed germination and post-germination growth (Lovegrove and Hooley, 2000; Ritchie and Gilroy, 1998b). In contrast, ABA suppresses GA induction of α-amylase gene transcription and blocks seed germination (Karssen et al., 1983; Koornneef and Karssen, 1994).

The GA signal is perceived by receptors such as GID1 (Ueguchi-Tanaka et al., 2005). The activated receptor triggers signal transduction events, including degradation of negative regulators by proteasomes (reviewed in Itoh et al., 2003; Sun and Gubler, 2004), elevation of cGMP (Penson et al., 1996) and cytoplasmic Ca²⁺ concentrations (Gilroy and Jones, 1992), induction of transcriptional activators, and expression of genes. cis-acting elements necessary and sufficient for the GA response of high-pI and low-pI α-amylase genes have been defined, and several transcription factors that bind to these elements have also been isolated (Rogers and Rogers, 1992; Tanida et al., 1994; Gubler et al., 1995; Kaneko et al., 2004; Lu et al., 2002; Rushton et al., 1995; Washio, 2003; Zentella et al., 2002; Zhang et al., 2004).

WRKY genes belong to a gene superfamily encoding transcription factors that are involved in the regulation of various biological processes, including pathogen defense, senescence and development (reviewed in Eulgem et al., 2000; Ulker and Somssich, 2004). Four of the six WRKY genes that are expressed in rice aleurone cells, OsWRKY24, -51, -71, and −72, are ABA-inducible (Xie et al., 2005). Furthermore, OsWRKY51 and OsWRKY71 share a similar expression pattern in response to GA and ABA in rice aleurone cells, and neither of them affects ABA induction of HVA22 transcription (Xie et al., 2005). Here, we report the co-expression of OsWRKY51 and OsWRKY71 in embryos and aleurone cells of rice seeds. OsWRKY51 and OsWRKY71 physically interact in the nuclei to synergistically suppress the induction of Amy32b by GA.

Results

OsWRKY51 and OsWRKY71 are co-expressed in seeds

In situ hybridization studies were performed with husked rice seeds treated with or without 10 μm uniconazole (a GA biosynthesis inhibitor) or 20 μm ABA for 30 h. DIG-labeled gene-specific antisense RNA probes were used to detect OsWRKY51 and OsWRKY71 mRNA in situ. In cross-sections of rice seeds treated without hormones, OsWRKY51 and OsWRKY71 signals were detected predominantly in the plumule, radicle and scutellum of the embryo, and the aleurone layer (Figure 1a,a′). OsWRKY51 and OsWRKY71 signals were enhanced by uniconazole (Figure 1b,b′) and ABA treatments (Figure 1c,c′), suggesting that the expression of these two genes in seeds is induced by ABA, but repressed by GA. This result is consistent with our previous observation that both OsWRKY51 and OsWRKY71 are upregulated by ABA and downregulated in response to GA treatments in rice aleurone cells (Xie et al., 2005).

Figure 1. OsWRKY51 and OsWRKY71 mRNA localization in imbibed rice seeds. Data for OsWRKY51 mRNA localization in embryo and aleurone tissues are shown in (a)–(d) and those for OsWRKY71 mRNA are shown in (a′)–(d′). Husked rice seeds were treated without (a,a′) or with 10 μm uniconazole (b,b′) or 20 μm ABA (c,c′) for 30 h. Cross-sections of seed were hybridized to the DIG-labeled antisense RNA probes (a–c and a′–c′). The DIG-labeled sense RNA probes were used as negative controls (d,d′). AL, aleurone layer; P, plumule; R, radicle.

OsWRKY51 suppresses GA induction of α-amylase gene expression

OsWRKY51 encodes a group II WRKY protein that contains one WRKY domain with a C₂H₂-type zinc finger motif (Figure S1a). Putative orthologs of OsWRKY51 have been found in wheat, barley, wild oat, maize and Arabidopsis (Figure S1b).

We have shown that OsWRKY71 suppresses GA induction of α-amylase expression (Zhang et al., 2004). Because OsWRKY51 and OsWRKY71 are co-expressed in embryos and aleurone layers of rice seeds and share a similar expression pattern in response to GA and ABA treatments (Figure 1), we studied whether they have a similar function in GA signaling in aleurone cells. The genomic fragment spanning from ATG to the stop codon of OsWRKY51 was cloned into an expression vector driven by the maize ubiquitin promoter (Figure 2a). Using the GA-inducible reporter construct, Amy32b-GUS (Lanahan et al., 1992), we showed that the exogenous GA₃ treatment resulted in a 42-fold induction of GUS activity over the control (Figure 2b). ABA completely blocked GA induction of Amy32b-GUS. Co-expression of UBI-OsWRKY51 decreased GA induction of Amy32b-GUS to twofold over the control. However, substituting a stop codon for codon of OsWRKY51 in the effector construct, which produced a truncated OsWRKY51 protein missing the WRKY domain and C-terminal end (OsWRKY51m), released 71% of the repression activity on GA induction (Figure 2b).

Figure 2. OsWRKY51 suppresses GA induction of the Amy32b promoter. (a) Schematic diagrams of the reporter and effector constructs used in the co-bombardment experiment. Rectangles denote exons and lines designate introns. Filled rectangles represent the WRKY domain. Hatched regions represent the GUS reporter gene. The 3′ deletion mutant (OsWRKY51m) encodes amino acids 1–201. (b) The effect of OsWRKY51 on GA induction of Amy32b-GUS. The reporter construct and the internal construct, UBI-LUC, were co-bombarded into barley de-embryo half-seeds without (−) or with (+) effector constructs. Bars indicate normalized GUS activities after 24 h of incubation of the bombarded half-seeds without (−) or with (+) 1 μm GA₃. Data are means ± SE of four replicates. (c) The dosage experiment was performed with a constant amount of the reporter construct (Amy32b-GUS) and various amounts of UBI-OsWRKY51 (0–100% relative to the reporter construct) in the presence of 1 μm GA₃. (d) The dosage experiment was performed with a constant amount of the reporter construct (HVA22-GUS) and various amounts of UBI-OsWRKY51 (0–100%) in the presence of 20 μm ABA.

The repression effect of OsWRKY51 was further confirmed by a dosage experiment, in which the amount of the reporter construct, Amy32b-GUS, was held constant, while that of the effector construct, UBI-OsWRKY51, varied from 0% to 100% relative to the molar amount of the reporter construct (Figure 2c). Upon increasing the amount of the effector construct, a gradual decline in expression of Amy32b-GUS was observed. Similar dosage experiments with the ABA inducible reporter construct HVA22-GUS (Shen and Ho, 1995) indicated that UBI-OsWRKY51 had little effect on ABA induction of HVA22-GUS (Figure 2d). These results suggest that OsWRKY51 is probably an ABA-inducible repressor of GA signaling.

OsWRKY51 and OsWRKY71 synergistically suppress GA induction of the Amy32b promoter

To study the combinatory effect of these two repressors, dosage experiments were performed for OsWRKY51 alone, OsWRKY71 alone, and OsWRKY51 with OsWRKY71 (Figure 3a). The results indicated that co-expression of UBI-OsWRKY51 at the 1% (effector/reporter) ratio had no suppression activity on the GA induction of Amy32b-GUS. Co-expression of 1%UBI-OsWRKY71 suppressed the GA induction of Amy32b-GUS to 91% of the control. Interestingly, co-expression of 1%UBI-OsWRKY51 plus 1%UBI-OsWRKY71 strongly reduced the GA induction of Amy32b-GUS to 25% of the control (Figure 3b). It is important to point out that this increased effect is unlikely to be due to twice as much effector DNA in the mixture. In fact, the reporter expression level resulting from 1%UBI-OsWRKY51 plus 1%UBI-OsWRKY71 was even lower than that for 2.5%, 5%, 7.5% or 10%UBI-OsWRKY51 or UBI-OsWRKY71 alone. These results suggest that the effect of OsWRKY51 and OsWRKY71 is not additive, but they synergistically act to suppress the GA induction of Amy32b.

Figure 3. OsWRKY51 and OsWRKY71 synergistically suppress GA induction of the Amy32b promoter. (a) Schematic diagrams of the reporter and effector constructs used in the co-bombardment experiment. The annotations are the same as in Figure 2. (b) The dosage experiment was carried out with a constant amount of the reporter construct (Amy32b-GUS) and varied amounts of the effector construct(s) in the presence of 1 μm GA₃ for 24 h. The relative amount of effector construct is indicated as a percentage compared to the amount of reporter. Each data point indicates mean GUS activities ±SE of four replicates.

Physical interactions of WRKY proteins are visualized in the nuclei of aleurone cells

OsWRKY71 is targeted to the nuclei of aleurone cells (Zhang et al., 2004). The UBI-GFP:OsWRKY51 fusion and the UBI-GFP control construct were introduced, respectively, into barley aleurone cells by bombardment. Confocal microscopic studies indicated that GFP fluorescence was detected in the entire cells bombarded with the UBI-GFP control (Figure 4a, panel I). However, the signal was only observed in the nuclei of the aleurone cells bombarded with UBI-GFP:OsWRKY51 (panel III). As controls, aleurone cell nuclei were stained with the red fluorescent dye, SYTO17 (panels II and IV). These results suggest that OsWRKY51 is also targeted to nuclei.

Figure 4. Visualization of interactions between WRKY proteins in aleurone cells by bimolecular fluorescence complementation. (a) Barley half-seeds were bombarded with either UBI-GFP (panels I and II) or UBI-GFP:OsWRKY51 (panels III and IV). After incubation for 24 h, the aleurone cells were stained with SYTO17 to localize nuclei (panels II, IV), followed by examination of GFP fluorescence (panels I and III). Arrows point to the same cell. The bars represent 20 μm. (b) Barley half-seeds were bombarded with either UBI-YFP or combination of constructs encoding indicated fusion proteins. YN, the fragment containing amino acid residues 1–154 of YFP; YC, the fragment containing amino acid residues 155–238 of YFP. After incubation at 24°C for 24 h, yellow fluorescence was observed through a confocal microscope.

We next studied whether OsWRKY51 and OsWRKY71 physically interact with each other in aleurone cells. The OsWRKY51 and OsWRKY71 coding sequences were fused in-frame with either that coding for the N-terminal fragment of YFP (YN) or that coding for the C-terminal fragment (YC). We tested the complementation between different combinations of fusion proteins by introducing the corresponding constructs into barley aleurone cells. Confocal microscopic studies showed that YFP fluorescence was detected in the entire cells bombarded with the UBI-YFP construct (Figure 4b, panel I). As expected, no fluorescence was detected when only one of fusion constructs was introduced into aleurone cells (data not shown). However, yellow fluorescence was observed in the nuclei of aleurone cells bombarded with the UBI-YN:OsWRKY71 and UBI-YC:OsWRKY71 constructs, suggesting that OsWRKY71 proteins interact in nuclei (panel II). Similarly, yellow fluorescence was detected in the nuclei of aleurone cells bombarded with either UBI-YN:OsWRKY51 and UBI-YC:OsWRKY71 (panel III), or UBI-YN:OsWRKY71 and UBI-YC:OsWRKY51 (panel IV), indicating that OsWRKY71 interacts with OsWRKY51 in the nuclei of aleurone cells. In contrast, no fluorescence was detected in aleurone cells bombarded with UBI-YN:OsWRKY51 and UBI-YC:OsWRKY51, suggesting that OsWRKY51 proteins might not physically interact with each other (panel V).

Interaction of OsWRKY51 and OsWRKY71 enhances the binding affinity of OsWRKY71 to the Amy32b promoter

We have previously shown that OsWRKY71 binds to the W boxes in the Amy32b promoter (Zhang et al., 2004). Because OsWRKY51 interacts with OsWRKY71 in nuclei (Figure 4b) to repress GA induction of the Amy32b promoter (Figure 3), we asked whether OsWRKY51 also binds the Amy32b promoter and whether the interaction of OsWRKY51 and OsWRKY71 has any effect on individual binding affinities. Electrophoretic mobility-shift assays (EMSAs) were performed with C-terminal His-tagged OsWRKY51 (51H) and N-terminal glutathione-S-transferase (GST)-tagged OsWRKY71 (G71). A DNA fragment that contains the Amy32b promoter region from −176 to −1 was labeled with α-³²P-dATP (Figure 5a). No binding signal was detected in reactions without protein (Figure 5b, lane 1) or with GST only (lane 2). Consistent with our previous report (Zhang et al., 2004), G71 bound to the Amy32b promoter specifically (lane 3). In the presence of 0.25 μm Zn²⁺, the binding affinity of G71 was significantly increased (lane 4) and an excess amount of unlabeled competitor (a DNA fragment from −171 to −111 which contains W boxes) competed with the in vitro binding (lanes 5 and 6).

Figure 5. Interaction of OsWRKY51 and OsWRKY71 enhances the binding affinity of OsWRKY71 to the Amy32b promoter. (a) The 371 bp probe (−176 to +79 plus the first intron of Amy32b) and 61 bp competitor (−171 to −111) used in EMSA in (b). Circles denote W boxes, the diamond represents the pyrimidine box, and the rectangle represents GARE in the Amy32b promoter. The DNA probe was end-labeled with α-³²P-dATP. (b) EMSA with recombinant OsWRKY51 and OsWRKY71 proteins and the ³²P-labeled DNA probe in the absence (−) or presence (+) of 0.25 μm Zn²⁺. Recombinant proteins, OsWRKY51:HIS (51H) and GST:OsWRKY71 (G71) were used in EMSA without or with an excess amount of competitor (20- or 200-fold). GST indicates the control binding reaction with the GST protein only. Binding reaction mixtures were separated on a 3.5% polyacrylamide gel. The gray arrow points to the binding signal of OsWRKY71; the black arrow points to the signal of the complex of OsWRKY51 and OsWRKY71. The free probe is indicated at the bottom. (c) The wild-type DNA fragment (−171 to −111 in the Amy32b promoter) was used as a probe in EMSA in (d), with unlabeled wild-type (WT) or mutant (M1–M4) competitors. Elements with a cross through them represent the mutated cis-acting elements in the competitor fragments. (d) EMSA with recombinant OsWRKY51 and OsWRKY71 proteins and the ³²P-labeled wild-type DNA probe in the presence of 0.25 μm Zn²⁺. Recombinant proteins, OsWRKY51:HIS (51H) and GST:OsWRKY71 (G71) were used in EMSA without or with an excess amount of competitor (20- or 200-fold). GST indicates the binding reaction with the GST protein only. Free probe is indicated at the bottom. The binding reaction mixtures were separated on a 5% polyacrylamide gel because the probe (71 bp) is much shorter compared to that (371 bp) used in (b). As a result, the super-shift signal for OsWRKY51/OsWRKY71 is less obvious than that in (b).

Unlike G71, 51H alone did not bind to the Amy32b promoter without (lane 7) or with Zn²⁺ (lane 8). In the absence of Zn²⁺, 51H and G71 together did not bind to Amy32b promoter either (lane 11). However, in the present of Zn²⁺, a strong signal was detected, and the complex was shifted higher than that obtained with G71 alone (compare lane 12 with lane 4). In addition, the excess amount of unlabeled DNA fragments competed with the binding of WRKY proteins (lanes 13 and 14). These results suggest that OsWRKY51 interacts with OsWRKY71 to bind the Amy32b promoter in a Zn²⁺-dependent manner.

To further demonstrate whether the OsWRKY51/OsWRKY71 complex specifically binds to two W boxes in the Amy32b promoter, a DNA fragment containing the W boxes and Pyr box of the Amy32b promoter (from −171 to −111) was used as a probe, and competition assays were carried out using the wild-type and mutant DNA fragments (Figure 5c). No binding signal was detected in the reactions without protein (Figure 5d, lane 1), with GST (lane 2), or with 51H only (lane 3). However, G71 bound to the wild-type Amy32b probe (lane 4), and an excess amount of unlabeled wild-type DNA fragment competed for this binding (lane 5). The fragment with only the second W box (M1, lane 6) competed for the binding of OsWRKY71 more efficiently than that with only the first W box (M2, lane 7), although both of them have a weaker competition than the wild-type fragment (WT). However, the fragment with both W boxes mutated (M3) showed little competition with the binding of G71 to the probe (lane 8). As a negative control, a mutation in Pyr box did not affect the binding of OsWRKY71 to W boxes (lane 9). Similar results were observed for the OsWRKY51/OsWRKY71 complex (lanes 10–20). The only difference was that much stronger signals were observed for the OsWRKY51/OsWRKY71 complex than OsWRKY71 alone. These results indicate that either OsWRKY71 only or the OsWRKY51/OsWRKY71 complex is capable of binding to either W box in the Amy32b promoter region, with a preference for the second one, and tandem-repeat W boxes significantly increase the binding affinity.

OsWRKY71 mutant proteins regulate the expression of Amy32b by modulating the binding affinities of OsWRKY51 and OsWRKY71 to the W boxes of the Amy32b promoter

To map the functional domains of OsWRKY71, wild-type and mutant derivatives (Figure 6a) of UBI-W71 (the full-length cDNA of OsWRKY71 driven by the ubiquitin promoter) were prepared and introduced into aleurone cells by co-bombardment. Co-expression of UBI-W71 decreased GA induction of the Amy32b promoter from 38-fold to threefold (Figure 6b). Surprisingly, mutations either in the WRKY motif (W71mW) or in conserved histidine residues of the zinc finger motif (W71mH) hardly affected the inhibitory activity on the GA induction even though these motifs are the hallmarks of the WRKY proteins. Even more surprisingly, the truncated OsWRKY71 that lacks of the WRKY domain and C-terminal region (W71m213) enhanced the expression of Amy32b-GUS, leading to a sixfold induction in the absence of GA. In the presence of GA, the induction level increased to 55-fold. However, the truncated OsWRKY71, which contains the WRKY domain but lacks the C-terminal region (G71m267), still had the repression activity although it was much lower than the wild-type.

Figure 6. Mutant OsWRKY71 proteins regulate the expression of Amy32b-GUS by modulating the binding affinity of OsWRKY51 and OsWRKY71 to the W boxes. (a) Schematic diagrams of the reporter and effector constructs used in the co-bombardment experiment. The full-length cDNA sequences of OsWRKY51 and OsWRKY71 were driven by the constitutive ubiquitin promoter in the effector constructs (UBI-W51 and UBI-W71). The mutations of OsWRKY71 are shown below the diagrams of the mutant constructs. (b) The reporter construct, and the internal control construct, UBI-LUC, were co-bombarded into barley de-embryo half-seeds without (−) or with (+) effector constructs. Unless indicated, equal molar amounts of effector and reporter were used. Bars indicate normalized GUS activities after 24 h of incubation of the bombarded half-seeds without (−) or with (+) GA₃. Data are means ± SE of four replicates. (c) Recombinant GST:OsWRKY71 (G71) and mutant OsWRKY71 proteins were used in EMSA with the ³²P-labeled DNA probe containing the promoter (−176 to −1), 5′ UTR, and the first intron of Amy32b (371 bp total). GST indicates the control binding reaction with the GST protein only; 10 μm 1,10-pananthroline was used as a chelator of the Zn²⁺ ions. (d) Recombinant OsWRKY51:HIS (51H), GST:OsWRKY71 (G71) and mutant OsWRKY71 proteins were used in EMSA with the ³²P-labeled DNA probe containing the promoter (−176 to −1), 5′ UTR, and the first intron of Amy32b (371 bp total). GST was added to ensure that the same total amount of protein was used in each reaction in lanes 1–4 and lanes 5–8 respectively. Free probe is indicated at the bottom. (e) The reporter construct and the internal control construct, UBI-LUC, were co-bombarded into barley de-embryo half-seeds without (−) or with (+) effector constructs. UBI-W51 and UBI-W71 effector constructs were used at a ratio of 1:10 without (−) or with (+) the UBI-W71m213 effector construct. Bars indicate normalized GUS activities after 24 h of incubation of the bombarded half-seeds without (−) or with (+) GA₃. Data are means ± SE of four replicates.

To understand how these mutant proteins regulate the expression of the Amy32b promoter, EMSA was carried out with GST-fused wild-type or mutant OsWRKY71 proteins. The DNA probe that contains the Amy32b promoter region from −176 to −1 was end-labeled with α-³²P-dATP. As shown in Figure 6(c), the wild-type OsWRKY71 bound to the Amy32b promoter (lane 2), and the binding was abolished in the presence of a Zn²⁺ chelator (1,10-phenanthroline) at a concentration of 10 mm (lane 3). Interestingly, although W71mW and W71mH retained significant levels of repression activity and W71m213 enhanced the expression of Amy32b-GUS, none of these proteins bound to the Amy32b promoter (lanes 4–6). In contrast, the binding activity of W71m267 to the probe was enhanced 13-fold (lane 7). The binding is specific as the signal was diminished by an excess amount of competitor (from −171 to −111 of the promoter, data not shown). These results suggest that (1) the region containing the WRKY domain (amino acids 213–266) of OsWRKY71 is necessary for DNA binding, and (2) the C-terminus (amino acids 267–348) is critical for the repressing activity of OsWRKY71 on GA induction of Amy32b.

The results described above raised an intriguing question: how do the mutant proteins, W71mW, W71mH and W71m213, modulate the expression of Amy32b without binding to its promoter region? One possible mechanism is that the mutant proteins could still bind to the promoter, although at a much lower affinity. Another possibility is that mutant proteins enhance or interfere with the binding of endogenous orthologs of OsWRKY51 and OsWRK71 in barley aleurone cells to the W boxes of the Amy32b promoter. The logic underlying this hypothesis is that wild-type barley aleurone cells were used in the functional assays throughout this study. Both barley and rice are cereal plants in which the GA and ABA signaling pathways are highly conserved. We used the barley aleurone layers because they are much easier to prepare and provide a much higher level of reporter gene expression (Zhang et al., 2004). Indeed, the putative barley ortholog of OsWRKY71 has been shown to bind to the promoter of Amy32bin vitro (Mare et al., 2004). To test this hypothesis, EMSA was carried out with the combination of wild-type proteins and mutant proteins by using the ³²P-labeled DNA probe that contains the −176 to −1 Amy32b promoter region (Figure 6d). As reported above, OsWRKY71 alone bound to the probe (lane 1) and OsWRKY51 increased the binding of OsWRKY71 by about 40% (lane 5). Similarly, the W71mW (lane 2) and W71mH (lane 3) mutant proteins, which are also unable to bind to the promoter, increased the binding activity of OsWRKY71 to the probe by four- and threefold respectively. A similar pattern was observed for the OsWRKY51/OsWRKY71 complex; the W71mW (lane 6) and W71mH (lane 7) mutant proteins further enhanced the binding activity of this complex to the probe. In contrast, addition of W71m213 decreased the binding signal by 35% for OsWRKY71 (compare lane 4 with lane 1) and by 77% for the OsWRKY51/OsWRKY71 complex (compare lane 8 with lane 5). This reduction is unlikely to be due to the increased amounts of proteins used in these two lanes. In fact, GST proteins were added in the reactions with OsWRKY71 (lane 1) or OsWRKY71 plus OsWRKY51 (lane 5) to ensure that the total amount of protein was the same for each reaction in the left panel (lanes 1–4) or each one in the right panel (lanes 5–8). These results indicate that W71mW and W71mH suppress the GA induction of Amy32b by increasing the in vivo binding of the barley orthologs of OsWRKY51 (Figure S1) and OsWRKY71 (Mare et al., 2004) to the W boxes of the Amy32b promoter in barley aleurone cells. In contrast, G71m213 enhances the GA induction of Amy32b by inhibiting the binding of the WRKY repressor proteins to the W boxes.

To further test the hypothesis, UBI-W71m213 was introduced into aleurone cells along with 10%UBI-W51 (the full-length cDNA of OsWRKY51 driven by the ubiquitin promoter) and 10%UBI-W71, alone or in combination. The results indicate that co-expression of 10%UBI-W71 or 10%UBI-W71 plus 10%UBI-W51 is sufficient to repress GA induction of Amy32b (Figure 6e). However, co-expression of UBI-W71m213 enhanced GA induction to 103-fold. W71m213 also reduced the inhibitory effect of OsWRKY51 plus OsWRKY71 by as much as 76% (compare column 2 with columns 4 and 7 respectively). Therefore, these functional assay data support the hypothesis that the W71m213 mutant protein interferes with the binding of OsWRKY51 and OsWRK71 to the Amy32b promoter, and therefore enhance the GA induction of Amy32b in aleurone cells.

OsWRKY51 and OsGAMYB functionally compete with each other to regulate the expression of Amy32b

OsGAMYB regulates GA induction of Amy32b (Gubler et al., 1997; Zhang et al., 2004), via GARE, an element 22 bp downstream of the W boxes. In previous studies, we have shown that OsWRKY71 blocks GA signaling by functionally interfering with OsGAMYB (Zhang et al., 2004). Dosage experiments were performed to study whether OsWRKY51 also has a similar effect on OsGAMYB induction of Amy32b. Varied amounts of UBI-OsWRKY51 (from 0% to 100% relative to the reporter construct) were introduced into aleurone cells along with a constant amount of UBI-OsGAMYB (100%) and Amy32b-GUS (100%). The OsGAMYB induction of Amy32b was gradually suppressed by increasing amounts of UBI-OsWRKY51 (Figure 7b). A similar dosage experiment was carried out with a fixed amount of UBI-OsWRKY51 (10%) and Amy32b-GUS (100%), and varied amounts of UBI-OsGAMYB (from 0% to 150%). As shown in Figure 7(b), the repression activity of OsWRKY51 was gradually overcome by increasing amounts of UBI-OsGAMYB. These results suggest that OsWRKY51 also regulates GA induction by interfering with the function of OsGAMYB. Thus, the ratio of repressors (such as OsWRKY51 and OsWRKY71) to the activator (e.g. OsGAMYB) seems to determine the expression level of the Amy32b promoter.

Figure 7. Functional antagonism between OsWRKY51 and OsGAMYB in regulating the expression of Amy32b-GUS. (a) Schematic diagrams of the reporter and effector constructs used in the co-bombardment experiment. The annotations are the same as in Figure 2. (b) The reporter construct (Amy32b-GUS) and the internal construct (UBI-LUC) were co-bombarded into barley de-embryo half-seeds with effector constructs (UBI-OsWRKY51 and UBI-OsGAMYB). The relative amount of effector construct is indicated as a percentage compared to the amount of reporter. The line with filled squares represents the dosage effect of OsWRKY51 (0–100%) on induction of UBI-Amy32b by 100% OsGAMYB; the line with open squares represents the dosage effect of OsGAMYB (0–150%) on repression of UBI-Amy32b by 10% OsWRKY51. Each data point indicates mean GUS activities ± SE of four replicates.

Discussion

For WRKY proteins with two WRKY domains (group I), the C-terminal domain is responsible for DNA binding, while the N-terminal domain might participate in the binding process by enhancing the binding affinity of the C-terminal domain (Eulgem et al., 1999; Ishiguro and Nakamura, 1994; de Pater et al., 1996). However, computational modeling based on the NMR solution structure of the C-terminal WRKY domain of the AtWRKY4 protein argues that both WRKY domains could bind to DNA, although the N-terminal one might have a much lower binding affinity (Yamasaki et al., 2005). Nevertheless, binding of the N-terminal domain has not been demonstrated in vivo or in vitro. Both OsWRKY51 and OsWRKY71 contain a single WRKY domain. Our EMSA studies demonstrated that OsWRKY71 behaves more like the DNA-binding C-terminal WRKY domain of the two-domain WRKY proteins, while OsWRKY51 behaves more like the non-DNA-binding N-terminal domain of those proteins (Figure 5). Therefore, it appears that this type of WRKY domain interaction could occur intramolecularly between two WRKY domains of the group I proteins or intermolecularly between a DNA-binding WRKY protein (such as OsWRKY71) and a non-DNA-binding WRKY protein (such as OsWRKY51).

Modulation of DNA-binding proteins by non-DNA-binding proteins has been reported for other gene families. For instance, group D basic helix-loop-helix (bHLH) proteins can heterodimerize with, and hence regulate the activities of, DNA-binding bHLH proteins (Ledent and Vervoort, 2001; Toledo-Ortiz et al., 2003). However, the group D proteins lack the DNA-binding basic domain, and heterodimerization between group D and other bHLH proteins blocks the DNA binding and transcriptional activation activities of the bound bHLH protein. This is different from OsWRKY51 because (1) it contains the WRKY domain, the potential DNA-binding region of the WRKY family, although it was unable to bind to DNA, and (2) its interaction with OsWRKY71 enhanced the binding affinity (Figures 5 and 6) and repressing activity (Figure 3) of OsWRKY71. In this sense, it is more like VIVIPAROUS1 (a B3 domain transcription factor), which can enhance the binding affinity of a basic leucine zipper protein (EmBP-1) to an ABA-regulated promoter and increase the activities of the EmBP-1 transcriptional activator (Hill et al., 1996).

This work reveals another node of the ABA and GA signaling cross-talk network. ABA downregulates many genes, especially those upregulated by GA. This effect is so drastic that ABA completely blocks GA-induced seed germination (reviewed in Lovegrove and Hooley, 2000). The cross-talk of GA and ABA signaling is mediated by secondary messengers. For example, application of PA to barley aleurone inhibits α-amylase production and induces an ABA-inducible amylase inhibitor and RAB protein expression, mimicking the effect of ABA (Ritchie and Gilroy, 1998a). Two regulators, VIVIPAROUS1 and SPINDLY, act as positive regulators of ABA signaling and negative regulators of GA signaling (Hoecker et al., 1999; Robertson et al., 1998). The ABA inhibition also involves kinases. For example, the ABA-inducible protein kinase PKABA1 is involved in the ABA suppression pathway (Gómez-Cadenas et al., 1999, 2001b). However, there is more than one pathway mediating the suppression of GA signaling by ABA, because the RNA interference of HvPKABA1 does not block the inhibitory effect of ABA on GA induction of α-amylase transcription (Zentella et al., 2002). Although it is tempting to suggest that the two ABA-inducible and GA-repressible WRKY genes presented in this work might represent components of this alternative suppression pathway, much work is needed to address the relationship between PKABA1 and OsWRKY51/OsWRKY71 in the ABA suppression pathway.

An intriguing question is how these two WRKY proteins modulate GA signaling. At least five cis-acting elements are essential for a high expression level of Amy32b: the O2S element that contains two TGAC cores, the pyrimidine box, GARE (GA response elements), the amylase box and DAE (downstream amylase element) (Gómez-Cadenas et al., 2001a; Lanahan et al., 1992). Repressors have been reported to bind to the O2S element (this work; Zhang et al., 2004), pyrimidine box (Mena et al., 2002; Washio, 2003), GARE (Raventós et al., 1998) and amylase box (Lu et al., 2002). In addition, activator proteins have been found to bind to the pyrimidine box (Isabel-LaMoneda et al., 2003), GARE (Gubler et al., 1995) and the amylase box (Lu et al., 2002). Therefore, we suggest that it is possible that each of the five cis-acting elements on the Amy32b promoter could be bound by at least a transactivator and a transrepressor, and that the ratio of the activators and repressors for individual cis-acting elements or the whole promoter complex controls the expression level of Amy32b (Zhang et al., 2004). Indeed, we have shown that the ratio of GAMYB activator and OsWRKY51/OsWRKY71 repressors determines the expression level of Amy32b-GUS (Figure 7) (Zhang et al., 2004). However, there is a difference for the two WRKY repressors: OsWRKY71 could bind to the promoter directly, whereas OsWRKY51 did not. In this sense, OsWRKY51 is similar to the non-DNA-binding repressor, KGM (kinase associated with GAMYB), which interacts with GAMYB and specifically represses the activity of an α-amylase promoter (Woodger et al., 2003). On the other hand, OsWRKY71 is like HRT (Hordeum repressor of transcription), which binds to GARE and suppresses gene expression (Raventós et al., 1998). We suggest that OsWRKY51/OsWRKY71 might block GA responses by interacting with the activation domain of GAMYB or by competing with GAMYB to interact with components of the basal transcription machinery. Of course, there are other possible modes. GAMYB and OsWRKY repressors might compete for interaction with the same co-transcription factors. Also, OsWRKY repressors might block the expression of α-amylase genes by enforcing the repression of KGM (Woodger et al., 2003) and HRT (Raventós et al., 1998), or interfering with binding of transcriptional activators to other cis-acting elements in α-amylase promoters.

In summary, OsWRKY51 and OsWRKY71 appear to mediate the cross-talk of GA and ABA signaling. ABA induces the expression of OsWRKY51 and OsWRKY71, and increases the ratio of OsWRKY51/OsWRKY71 repressors to GAMYB activator. The repressors bind the W boxes, thus exerting an antagonistic effect against GA induction of Amy32b. In contrast, GA induces GAMYB, but suppresses OsWRKY51 and OsWRKY71. The ratio of OsWRKY51/OsWRKY71 to GAMYB is low so that GAMYB competes against the inhibitory effect of OsWRKY51 and OsWRKY71, leading to a high level of the α-amylase gene expression. Obviously, this is an over-simplified model because it does not consider other activators such as OsDOF3 (Washio, 2003), RAMY (Peng et al., 2004) and OsMYBS1/OsMYBS2 (Lu et al., 2002), and other repressors such as KGM (Woodger et al., 2003) and HRT (Raventós et al., 1998). Nevertheless, this model gives us new insights about how multiple transcription factors coordinate to fine tune the Amy32b expression that is triggered by different signals. Much work needs to be done regarding the relationships among OsWRKY51/OsWRKY71, other repressors and activators mentioned above, and other components of the GA and ABA signaling networks.

Experimental procedures

Plant materials and incubation conditions

Rice (Oryza sativa L. cv. Nipponbare) seeds were used in this study. For mRNA in situ hybridization, husked rice seeds were treated in water, 10 μm uniconazole (Wako, Osaka, Japan) or 20 μm ABA (Sigma, St Louis, MO, USA) for 30 h respectively. Barley (Hordeum vulgare cv. Himalaya) seeds (1998 harvest) were purchased from Washington State University (Pullman, WA, USA).

In situ hybridization

The gene-specific OsWRKY51 fragment was amplified by PCR using primers 5′-CAGCAGCAGCAGGTGGAC-3′ and 5′-GCCGCTGCCACTGCTCCAA-3′ and the plasmid UBI-W51 as the template. For the OsWRKY71 fragment, PCR was performed with primers 5′-AGAACAGCGACGGCTCCGGCAAG-3′ and 5′-GATCGATCGAACTCCGCCATGG-3′ and the genomic DNA isolated from seedlings. The procedure to prepare RNA probes is as follows: the PCR fragments were cloned into the pBluescript SK(+) plasmid, and the clones were confirmed by sequencing. The plasmids were either treated with BamHI and transcribed with T7 RNA polymerase (Stratagene, Cedar Creek, TX, USA) to prepare sense probes or digested with SalI and transcribed with T3 RNA polymerase (Stratagene) to prepare antisense probes. The procedure of in situ hybridization was performed as described previously (Kouchi and Hata, 1993).

Effector construct preparations

The coding region sequences that contain exons and introns from the start codon to the stop codon of OsWRKY51, OsWRKY71 or OsGAMYB were amplified by PCR, and cloned into the expression vector UBI promoter-linker-nos terminator, generating UBI-OsWRKY51 and UBI-OsWRKY71. The full-length cDNA of OsWRKY51 and OsWRKY71 was amplified from total RNA of rice aleurone cells by RT-PCR, and cloned into the AscI site of the expression vector using the following primers: 5′-ATCCGGCGCGCCTTGATGATTACCATGGATC-3′ and 5′-TTGGCGCGCCGGGGTATCTCATGCGAGCG-3′ for preparation of UBI-W51, and 5′-TTGGCGCGCCATGGATCCGTGGATTAGC-3′ and 5′-ATAGGCGCGCCTCAATCCTTGGTCGGCGAG-3′ for preparation of UBI-W71. Oligonucleotide-directed mutations (Kunkel et al., 1987) for UBI-OsWRKY51 and UBI-W71 were performed with the following primers: 5′-GGAGTGCGCGTGGTGAATTCACTTGCTCTTGTGGCCG-3′ for UBI-OsWRKY51m, 5′-GGTGACCTTCTGACCATCGGCGCGCCTTTGGTACCCATCTTTCAC-3′ for UBI-W71mW, and 5′-CGGCTGGCCGGAATTCTGCTCCCCCTCGTACGTCGC-3′ for UBI-W71mH. The 3′ deletion mutants of UBI-W71 were amplified by PCR using the following primers: 5′-TTGGCGCGCCATGGATCCGTGGATTAGC-3′ and 5′-TATAGGCGCGCCTCATGGGCAGGGGTTGTCCTTG-3′ for UBI-W71m213, and 5′-TTGGCGCGCCATGGATCCGTGGATTAGC-3′ and 5′-TATAGGCGCGCCTCAGCCGTCGCTGTTCTGCG-3′ for UBI-W71m267. The PCR products were then cloned into the AscI site of the UBI promoter-linker-nos terminator construct.

Particle bombardment and transient expression assays

The procedure of transient expression experiments with the barley aleurone system by particle bombardment was performed as described previously (Lanahan et al., 1992). The amount of effector construct used in the transient expression experiments was indicated as a percentage. For example, 10% of the effector indicates that the molar ratio of the effector construct to the reporter construct was 10%. Unless otherwise indicated, the molar ratio of the effector construct to the reporter construct was 100% in transient expression experiments. After bombardment, four half-seeds were incubated in 4 ml of the imbibing solution containing 50 μg ml⁻¹ chloramphenicol without or with 1 μm GA₃ or 20 μm ABA in small (6 cm diameter) petri dishes at 24°C for 24 h. All experiments consisted of four replicates and each experiment was repeated at least twice with similar results. GUS activity was normalized in every independent transformation relative to the luciferase activity.

Subcellular localization and bimolecular fluorescence complementation

The OsWRKY51 coding region in UBI-OsWRKY51 was amplified by PCR using primers 5′-ATCCGGCGCGCCTTGATGATTACCATGGATC-3′ and 5′-TTGGCGCGCCGACAAGCTAAATTCTCACTC-3′, and inserted into the AscI site of UBI-GFP (Zhang et al., 2004) to generate UBI-GFP:OsWRKY51. The YFP coding region was amplified by PCR using the primers 5′-TGGTACCATGGTGAGCAAGGGCG-3′ and 5′-ACGGTACCGTACAGCTCGTCCATG-3′, and inserted into the KpnI site of the UBI promoter-linker-nos terminator, to generate the UBI-YFP construct. The sequences encoding amino acid residues 1–154 (YN) and 155–238 (YC) of YFP were amplified from pQE801-YN and pQE80-bFosYC (Tsuchisaka and Theologis, 2004) and cloned into the KpnI site in the UBI promoter-linker-nos terminator using the following primers: 5′-ATGGTACCATGGTGAGCAAGGGCG-3′ and 5′-CTAGGTACCTAGAGCATGCTTAAGAGATC-3′ for UBI-YN, and 5′-ATGGTACCATGGACAAGCAGAAGAACG-3′ and 5′-TCCGGTACCGGATCCAGATCCAC-3′ for UBI-YC. The OsWRKY51 coding region was isolated from UBI-OsWRKY51 and inserted into the AscI site in UBI-YN and UBI-YC to produce UBI-YN:OsWRKY51 and UBI-YC:OsWRKY51. The OsWRKY71 cDNA was isolated from UBI-W71 and inserted into the AscI site in UBI-YN and UBI-YC to produce UBI-YN:OsWRKY71 and UBI-YC:OsWRKY71.

After incubation at 24°C for 24 h, the aleurone layers were peeled from the bombarded barley half-seeds. Confocal microscopic observations of the GFP fluorescence and nucleus staining with SYTO17 were performed as described previously (Zhang et al., 2004). The yellow fluorescence was observed using a laser scanning microscope LSM 510 with 514 nm excitation and a LP530 filter (Carl Zeiss, Inc., Göttingen, Germany). The acquired images were processed using Paint Shop Pro 7 (Jasc Software, Eden Prairie, MN, USA).

Production of recombinant proteins in Escherichia coli

The full-length cDNA sequence of OsWRKY71 was cloned into the AscI site of the modified pGEX-KG (Zhang et al., 2004) to generate GST:W71. For mutant protein production, derivatives of OsWRKY71 were subcloned into the AscI site of the modified pGEX-KG. The fusion constructs were then introduced into BL-21 (DE3) pLysS E. coli (Novagen, Madison, WI, USA). The GST fusion proteins were purified using glutathione–agarose beads (Zhang et al., 2004).

The full-length cDNA sequence of OsWRKY51 was amplified from UBI-W51 and cloned into the EcoRI site of the pET21b vector (Novagen) using the following primers: 5′-TCGAATTCGATGATTACCATGGATCTGATG-3′ and 5′-GCGAATTCGCGAGCGGCAGCG-3′. E. coli strain Origami B DE3 (Novagen) was transformed with this expression plasmid. A fresh colony was inoculated in 2 × yeast tryptone (YT) liquid medium containing 100 μg ml⁻¹ carbenicillin and was grown at 37°C till the middle log phase. Expression of His-tagged OsWRKY51 proteins was induced by the addition of 1 mm isopropyl β-d-thiogalactoside at 25°C for 5 h. The recombinant protein was purified using Ni-agarose affinity chromatography (Qiagen, Valencia, CA, USA) under native conditions by following the manufacturer's instructions.

Electrophoretic mobility shift assay

After digestion with NcoI, a 371 bp DNA fragment that contains the Amy32b promoter region from −176 to −1 and its first intron was isolated from the Amy32b-GUS (Lanahan et al., 1992). The DNA probe was labeled with α-³²P-dATP by Klenow fill-in reactions. The sequences of synthetic oligonucleotides used as competitors were as described previously (Zhang et al., 2004). Unless otherwise indicated, binding reactions contained 1.5 ng of probe, 1 μg of poly-(dIdC), 10 mm Tris–HCl (pH 7.6), 50 mm KCl, 0.25 μm ZnCl₂, 10% glycerol. Equal amounts of recombinant proteins (0.5 μg for each) were added into reactions and incubated at room temperature for 30 min with labeled DNA probes in the absence or presence of competitors. After incubation, the reactions described in Figure 5(d) were analyzed by electrophoresis on a 5% polyacrylamide gel for 5 h, while all other reactions were resolved on 3.5% polyacrylamide gels in 0.5 × TBE buffer (45 mm Tris, 45 mm boric acid, 1 mm EDTA) for 5 h. The signals were scanned with a Typhoon 9400 phosphor imager (Amersham Biosciences, Piscataway, NJ, USA) and quantified with ImageQuant software (version 5.2, Amersham Biosciences).

Acknowledgements

We thank Drs Andrew Andres, Jose Casaretto, Allen Gibbs, David Ho, Eduardo Robleto, Athanasios Theologis, Kenji Washio and Helen Wing for advice or contributions to the improvement of this paper. This work is supported by UNLV PIA, NIH BRIN (P20 RR16464) and a NSF EPSCoR grant to Q.J. Shen, Z.L. Zhang, and Z. Xie. G. Yang was supported by a Japanese BRAIN Postdoctoral Fellowship.